fixpafandomcom-20200216-history
General Info on polling
Robert Meadow- Decision Research, Founder and President. Research and Polling. Polls and focus groups guide many campaign decisions. Why take a poll? To get a sense of the public's knowledge and views. Asking your friends may give you a distorted picture. Public Opinion Research tells you what others think. What issues move people? How can you get your candidate from point A to point B? How do you match up your candidate's positons, values, etc with those of the electorate? A pollster can't change a candidate's view on an issue, but the candidate may decide to emphasize or deemphasize a part of his or her beliefs. Candidate may skip over issues and topics not of interest to the voters. People 30-50 years old typically have school age kids. They are more interested in education than 65 year olds. 65 year old is interested in social security. Candidates can emphasize different aspects of their platform with different voter groups, without lying. Polls help you to match the concerns and interests of the public with the concerns and interests of your ccandidate. Also find out the vocabulary that is used by different segments of the voter population. Candidates must be consistent. You can't be for vouchers here and against vouchers there. Polls give you empirical (quantifiable) evidence of public's concerns. Polls help guide the campaign's message development. Every campaign needs a core message that resonates with the electorate. Pollsters help craft the campaign message, which will lead to a majority voting for you. Must use the right vocabulary. Talking points and issues of candidate must move people. Use polls to identify targets. Suppose 40%D 40%R and 20% undecided. You want to know more about the 20% persuadable (undedcided). The issues of the undecideds may be different. Pollsters try to identify the undecideds, the swing voters. Find out the demographics of the swing voters. Ex. theyh may be 50-70 yrs old females on a fixed income, etc. Polling helps to find the swing voters. Base + swing voters = an electoral majority. Polling is used for message development. To find out which messages work with different sugcategories of the persuadable voters. Iraq, soc sec, school choice, educ, etc. One of these issues may move various subgroups of the electorate. Build a winning coalition this way. Polls vs. focus groups. Polling is scientifically-designed and implemented by professionals. Souns statistical principles guide polling process. Zogby National Sample may interview only 1,200 people, yet be statistically valid. You need a strategic understanding of the voters. What issues motivate them? Is the general mood for change or for the status quo? Is the country on the right or wrong track? What characteristics are the voters looking for in the candidate's bio? (emphasize war record, voting record, education?) Find out what kind of background information on the candidate would be persuasive. Find out which messages will work. Pollsters may test many themes and messages in their interviews to find one that resonates the best. They ask, 'if you knew that candidate X had this view _______, would you vote for him? More likely, somewhat likely, no diff, less likely, much less likely, etc. Ask 15 of these and rank them in order of responsiveness. Persuadables may have a different #1 issue than that of the general electorate. There are limited campaign resources so you must locate, contact and persuade the swing voters. Polling isn't an afterthought. It is at the core of the campaign strategy. You might have a great idea for a funny TV ad but the public (or the swing voters) may not be interested in that issue. So make another ad. Campaign team works with pollsters to design a message, which is than communicated via the media. Campaign strategies come from polling surveys. TV and radio and newspaper polls are not scientific. They mostly focus on the horse race aspect. Poll may be held one year before the election. How do you get your candidate from where he is now, to where you want him to be on election day. Candidate may only be at 25% among retired men. Winning scenario needs candidate to be at 42% among them by election day. What messages will get my candidate from where he is now to a win on election day? Identify subcomponents of the electorate who will move between now and election day. TV and newspaper polls often don't poll the likely electorate. They just call people out of the phone book. Political pollsters interview the likely electorate. The politician's friends are not representative of the likely electorate. They have inside knowledge and special interests in the race. They aren't the same as persuadable voters, who might not think about politics very much. So talk to experts for policy research, but not for polling purposes. Winning campaigns need to ask dozens of questions, not just the 'horse race' question. Polling interviews are done by professional interviewers for consistency, objectivity and neutrality. Campaign volunteers are best used elsewhere to persuade voters. Volunteer pollsters might try to influence the person being interviewed, and this biases the results. You don't want to deceive yourself. Three main types of polls. Benchmark, Trend polls, and tracking polls. Benchmark poll. 17-20 minutes. For early research and message development. Done 3 months to one year before election. Gives you the lay of the land. Use this to plan the campaign strategy and message. Trend polls (10min) used to update the campaign. Things may have changed since the benchmark poll. You might want to subtly change your message. See the effects on new information on the race. Tracking polls (3-5 min). Use this to monitor the effectiveness of the voter contact program. Find out if your message is getting across via TV and direct mail. Are you moving in the right direction? Are you picking up support? It's a waste of money to do a tracking poll the day before the election because there's no time to change anything anyhow. Test different issues and themes in a benchmark poll. Interviewees get fatigue after 20 minutes of an interview. Lose all data in a terminated interview. Politics is not central to soccer moms. Benchmark poll is most important poll. Do Benchmark if you can only do one poll. 3 month long media campaign. Do a trend poll after one month. See if your message is getting across. Use tracking polls to adjust your field voter contact program. Where to deploy resources in final week? Can't change your TV ads usually in the last week of campaign, but you can redeploy your volunteers. Sampling concepts. Two factors set the cost. Interview length and sample size. Error is based upon sample size, not size of the population. 1200 people in a national sample is enough. Sampling error is +/- 3%. 1200 interviews. 600men. 600 women. Divide country in N,S,E,W. So only 150 eastern women in the east. So only 75 female GOP in the east. Divide into 3 age categories and there are only 25 GOP women between 20-45 in the east. It is hard to generalize from only 25 people. A state might have 5 media markets. May need a bigger sample size, or you may need to over-represent a subgroup in your poll. Need 100 people in a subgroup to be confident in the reaction to a proposed campaign theme. Hard to generalize from a small subgroup, so oversample that one demographic. Ex. interview and extra 100 African American men. Do 1,000 + 100 = 1,100 interviews. A 1cc blood sample is representative of all of your blood, which telly you a lot about the whole body. A vat of soup that is well-stirred - a spoonful will tell you what it all tastes like, as long as it has been stirred. People in the same demographic will think similarly. So samples work and are statistically valid. A sample size of 1,200 out of 240 million is okay. But take more samples if you divide into subgroups. Age, religion, gender, pol party. Do research before taking the benchmark poll. Look at the following: Defensive and opposition research (must be objective). You want no surprises during he campaign. Figure out your responses beforehand. Look at press clippings, campaign brochures, strengths and vulnerabilities of both candidates, different issues, different potential message carriers (endorsements). Look at voting records of candidate and your opponent. Want to anticipate the opponent's moves so you know which ones to respond to, and which charges aren't going anywhere. Ignore the one-day stories and stay on your message. Find out ahead of time which endorsements carry the most weight with which demographic groups. What to ask in a poll? Issues, priorities, hot buttons Name recognition and favorability Balanced arguments for/against your candidate's positions. Trial heats/ Bio trial heats. endorsers/ message carriers. Get issues ranked from most important to least impt. Candidate may talk about campaign finance reform, but no one really cares. Persuadable voters may have their own #1 issue. Persuadables may care #1 about healthcae, but it might be the #6 issue overall. Target healthcare info to the persuadables. Candidate with high unfavorables may have problems. Sometimes city council candidates may only have 25% voter recognition. Senators have 95%. Prez has 99%. Identify which arguments are more likely to get you more votes. Also test reasons to vote for or against a candidate. You will know which attacks to ignore and which to respond to. People who change their opinions during the interview are probably undecideds. So ask a few times the 'trial heat question'. Who would you vote for right now? Supporters say this. Opponents say that. Which do you believe? Message box. What should you say about yourself? What should you say about your opponent? What will your opponent say about you? What will he say about himself? Identify persuadable voters. Typically 15-25% might be persuadable. In small local races, up to 50% can be persuadable. Fielding (interviewing). Well trained. not for volunteers. Methods are good. Calls are audited. Vary the time of day and the days of the week that they are called. Limit 3-6 days to finish a survey, to get a snapshot. Volunteers would take a month to do a 3 day survey. Clients want to listen to the interviews, but this is like watching sasusage being made. But they are paying, and they are curious. Don't mispronounce their names! Call during weekdays and get only seniors, homemakers, unemployed. Usually call 5-9 weekdays and 1-9 on weekends to get a good mix of people. Tracking polls are used to make mid course corrections. To identify effectiveness of media programs and ads. To determine if more resources are needed. Use tracking polls only when it is still possible to adjust tactics. But often the candidate is curious. One shot tracking poll may cost $5,000 to $10,000. Continuous tracking is expensive. Snapshot tracking - do it once. Waves- do one every couple of weeks. Rolling track- 200 interviews every night in last 30 days of the election. Drop old data and include the new data. Use most recent 7 days. Polling reports. Topline- ex. 30% want to raise the soc sec age. Cross tabulated across subgroups. Breaks aggregate data into various demographics. Statistical modeling may help you to find out which variables identigy supporters of a particular candidate. Some subgroups are larger than others. Focus Groups. Get 8-12 people in a conference room with one way mirror for 2 hours. Have a directed conversation. Get qualitative info. Does not represent the electorate. You can probe in depth. Show them ads and direct mail pieces. Then take them away and see what they remember 5minutes later. Focus group could be all undecided, or all women 40-60 YO. Show them an ad, see if they like it. Hear real people with their real vocabulary. You get verbal and nonverbal cues. Group dynamics - one loudmouth can ruin the whole session. So thank him, pay him, ask him to leave. Groups often try to reach consensus. There is an artificial intensity to the discussions. Also who shows up for focus groups? they get $60 to show up. Focus groups cost $5-10 thousand per group. Benchmark poll costs $18-40 thousand Tracking polls cost $5-50 thousand Usually 5-7% of campaign budget should be for polling and focus groups. But 80% is spent on voter contact (ads and brochures). So polling makes sure that the ads are going to register with the electorate. Refusal rate- How to compensate for hangups (20-30%)? Telemarketers get a much higher hangup rate than political pollsters. Make sure the distribution of people called is same as that of likely voters. 20% of electorate is over 65. You can overweight or underweight respondents in various demographic candidates to arrive at the expected electorate. Under 30- may only have cellphones. They are harder to reach. The best predictor of your attitudes are those of your neighbors, statistically speaking. Good pollsters pay lots of attention to sampling. Polls give you the intensity -- very likely, somewhat likely, neutral, etc. Intensity is quantified. Open ended questions are better in focus groups than in polls. The benchmark poll is the most essential. If a state rep race only has one TV commercial and it's running the last week before the election, then why do a tracking poll? You can't change anything. Tracking is for tactical changes. Can test an ad in a focus group to see if it will be persuasive. Focus groups are good if the candidates have strong, distinctive personalities. Robert Meadow was an academic who did polling and political consulting on the side. Pollsters come from three groups: 1. formal academics 2. Market researchers 3. Political operatives Many states have this info on file: Date of birth, date of registration, race (some states), party, voting history, address. Match your sample to the official voter file. Use census data to estimate education, income, etc. You may have to make an educated guess. Push polling is actually advocacy phoning. Push polls are different from survey research. Advocacy calls can be much shorter than survey research. Don't mix survey polling with advocacy phoning. Use surveys to shape your message, then use advocacy phoning later onto persuade people. Partial interviews are still usable, if you have enough demographic informatilon. Focus group may ask a loudmouth to leave. Thank them, pay them, let them go home. Opponent launches an attack on you. How to respond? If it is a one day story, just stay on your message. You won't have time to wait for a poll, so test the responses much earlier. the end.